DrBERT-CASM2 / tokenizer_config.json
olivierb's picture
initial commit
e5844c7
raw
history blame contribute delete
415 Bytes
{
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": true,
"mask_token": "[MASK]",
"model_max_length": 128,
"name_or_path": "Dr-BERT/DrBERT-4GB-CP-PubMedBERT",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": null,
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]"
}